Im morgigen #DigitalHistoryOFK stellt Marten Düring (C2DH) das Projekt #Impresso vor, das nicht nur eine Vielzahl von historischen Medien wie Zeitungen & Radioquellen sprach- & ländergrenzenübergreifend recherchierbar macht, sondern auch über die Verknüpfung mit #MachineLearning-Methoden datengetriebene Forschung & neue Perspektiven fördert.
I continue to be saddened by the lack of a good #data, #machinelearning or #AI community on the #Fediverse ... the conversation just really seems to be all on the birdsite! Or am I missing something obvious? I follow all the hashtags etc, but really feels a little thin on the ground.
Abstract:
Recently, there has been considerable interest in large language models: machine learning systems which produce human-like text and dialogue. Applications of these systems have been plagued by persistent inaccuracies in their output; these are often called “AI hallucinations”. We argue that these falsehoods, and the overall activity of large language models, is better understood as bullshit in the sense explored by Frankfurt (On Bullshit, Princeton, 2005): the models are in an important way indifferent to the truth of their outputs. We distinguish two ways in which the models can be said to be bullshitters, and argue that they clearly meet at least one of these definitions. We further argue that describing AI misrepresentations as bullshit is both a more useful and more accurate way of predicting and discussing the behaviour of these systems.
GPT-4o (“o” for “omni”) is a step towards much more natural human-computer interaction—it accepts as input any combination of text, audio, and image and generates any combination of text, audio, and image outputs.
Die 1. Sitzung ist wieder ausgewählten Abschlussarbeiten gewidmet:
Julia Pabst untersucht, wie #MachineLearning in der #Epigraphik zur Identifikation von Wappen & Inschriften eingesetzt werden kann & Lukas Germann widmet sich den Herausforderungen der Analyse von #Twitter-Daten.
Now, that being said, here's some GOOD uses of AI:
Correct grammar/spelling/word usage
Summarizing long form text
Suggestions for a wide variety of things.
Searching (Fuck you google)
DnD gamerunning, either assisted or as an actual DM (this one I am probably misrepresenting, unfortunately. I am not an actual DnD player. So I guess I am wishing this😬)
Secrets of Machine Learning: How It Works and What It Means for You by Tom Kohn, 2024
Cutting through the mass of technical literature on machine learning and AI and the plethora of fear-mongering books on the rise of killer robots, Secrets of Machine Learning offers a clear-sighted explanation for the informed reader of what this new technology is, what it does, how it works, and why it's so important.
The misleading readout, however, is not unusual and exposes weaknesses in the AI-generated software that many believe still needs fine-tuning.
👏 You 👏 can't 👏 fix 👏 accuracy 👏 problems 👏 with 👏 machine 👏 learning 👏 language 👏 models
They're a fundamental aspect of the technology. It's magnetic poetry with extra steps. Not an answer machine.
In fact, such errors have sparked a bigger backlash worldwide, with a rise in the number of lawsuits over poor accessibility to websites for disabled people.
This will not end. The answer is to hire actual people to provide actual accessibility. Sowwy CEOs :( :( :( :( :(
So annoying that they will not, under any circumstance, understand any of this. All that can happen is the financial loss and legal liability finally become too great.
Societal Impacts of Artificial Intelligence and Machine Learning by Carlo Lipizzi
This book goes beyond the current hype of expectations generated by the news on artificial intelligence and machine learning by analyzing realistic expectations for society, its limitations, and possible future scenarios for the use of this technology in our current society.
sounds like the much heralded job of the future, "prompt engineer" is no longer needed 😅
"Battle and his collaborators found that in almost every case, this automatically [AI generated] generated prompt did better than the best prompt found through trial-and-error. And, the process was much faster, a couple of hours rather than several days of searching."
Quirks of machine learning interpretations of images. The dataset interrogation/inspection of the original 1st edition Advanced Dungeons&Dragons Players Handbook cover, correctly grabs there is a giant monster like face in the background and that the cover is about D&D. But when you take all the phrases and ask it to push out an image based on what it thinks in the image. It makes a more modern godzilla monster, and it actually makes a group of people playing a rpg at a table beneath the monster.
I added the trade dress manually to the image, for effect.
Imagine the field day the Satanic Panic group would have had if this were the cover of the PHB in the 80s.
🥁 In the final session of our #DigitalHistoryOFK for this semester, we welcome Thea Sommerschield (University of Nottingham), who will introduce us to the current trends, challenges & future prospects in the field of #MachineLearning and #generativeAI for the study of Ancient Languages and media (from cuneiform to carbonised papyri). Not to be missed!
The terrible human toll in Gaza has many causes.
A chilling investigation by +972 highlights efficiency:
An engineer: “When a 3-year-old girl is killed in a home in Gaza, it’s because someone in the army decided it wasn’t a big deal for her to be killed.”
An AI outputs "100 targets a day". Like a factory with murder delivery:
"According to the investigation, another reason for the large number of targets, and the extensive harm to civilian life in Gaza, is the widespread use of a system called “Habsora” (“The Gospel”), which is largely built on artificial intelligence and can “generate” targets almost automatically at a rate that far exceeds what was previously possible. This AI system, as described by a former intelligence officer, essentially facilitates a “mass assassination factory.”"
"The third is “power targets,” which includes high-rises and residential towers in the heart of cities, and public buildings such as universities, banks, and government offices."
The sources said that the approval to automatically adopt #Lavender’s kill lists, which had previously been used only as an auxiliary tool, was granted about two weeks into the war, after intelligence personnel “manually” checked the accuracy of a random sample of several hundred targets selected by the #AI system. When that sample found that Lavender’s results had reached 90 percent accuracy in identifying an individual’s affiliation with Hamas, the army authorized the sweeping use of the system. From that moment, if Lavender decided an individual was a militant in Hamas, the sources were essentially asked to treat that as an order.
“Still, I found them more ethical than the targets that we bombed just for ‘deterrence’ — highrises that are evacuated and toppled just to cause destruction.”
The sources said that the approval to automatically adopt #Lavender’s kill lists, which had previously been used only as an auxiliary tool, was granted about two weeks into the war, after intelligence personnel “manually” checked the accuracy of a random sample of several hundred targets selected by the #AI system. When that sample found that Lavender’s results had reached 90 percent accuracy in identifying an individual’s affiliation with Hamas, the army authorized the sweeping use of the system. From that moment, if Lavender decided an individual was a militant in Hamas, the sources were essentially asked to treat that as an order.
“Still, I found them more ethical than the targets that we bombed just for ‘deterrence’ — highrises that are evacuated and toppled just to cause destruction.”